No one is doing that, though. What cosmologists do is parameterize the statistics of the DM distribution. That's one or two parameters. Then we compare observations to simulations to determine how likely the observed distribution is given the statistical properties. For example, a few galaxies with almost no dark matter would be expected due to the dynamics of clusters and galaxies. You could in principle calculate how often that should be the case, and if we were to observe it much more often than we should then there would be a problem with DM. No one is suggesting that the DM distribution can assume any arbitrary shape.
No, variation in galaxy properties is an output, not an input, of the model.
You could decide to quantify and catalog different galaxies with one or more parameters that describe their properties. You could then compare whether that catalog is statistically consistent with the output of the model (and must take into account all uncertainties in the model and the observations).
By analogy, you can measure that different people have different heights, but it does not mean that the specific height of each individual person is a unique input parameter in any fundamental model of biology.
That's what I mean by "it's a per-galaxy parameter". For each galaxy, to explain the behavior of that galaxy, you're saying "it must have X amount of dark matter".
The focus on "per-galaxy parameters" is like expecting to be able to predict how tall Tom Cruise should be after reading a textbook on the theory of evolution.
I know you are already aware, but that is literally the entire premise of MOND -- the M is for "MOdified Newtonian Dynamics".